VISION-RISK: Vision-Language Model for Risk Assessment in Explainable Autonomous Driving Systems

EasyChair Preprint 16014

Abstract

The advancements of autonomous driving systems hinge on their ability to navigate complex environments while ensuring safety and transparency. The lack of explainability in the current technologies - the ability to provide clear, human-readable justifications for actions - undermines trust, complicates validation and hinders widespread adoption. In this paper, we introduce VISION-RISK, a vision-language model (VLM) designed for risk assessmenet and explainability in autonomous driving using a lightweight architecture, optimized for deployment on edge devices. To train the model, we developed a custom dataset combining real-world driving scenarios from Honda Driving Dataset (HDD) and extreme high-risk cases from Crash1500, augumented with synthetic annotations using Dolphins and refined via DeepSeek V3. VISION-RISK stands out through three key characteristics: the integration of danger level classification with natural language explanation generation, a lightweight architecture optimized for deployment on resource-constrained devices, and a strong emphasis on interpretability and safety to enhance trust in autonomous systems.

Keyphrases: Explainability, VLM, autonomous driving

@booklet{EasyChair:16014,
  author    = {Andrei Bogdan Constantin and Sebastian-Antonio Toma and Vlad Negru and Camelia Lemnaru and Rodica Potolea},
  title     = {VISION-RISK: Vision-Language Model for Risk Assessment in Explainable Autonomous Driving Systems},
  howpublished = {EasyChair Preprint 16014},
  year      = {EasyChair, 2026}}